A framework for formalization and characterization of simulation performance

Abstract

Researchers have lamented that the lack of adequate performance evaluation framework hinders the widespread adoption of parallel discrete-event simulation. This thesis proposes a new framework for analyzing the performance of simulation that consists of two parts: formalization and characterization. The formalization of various simulation event orderings provides a theoretical foundation for conducting performance analysis and facilitates the understanding of the relationship of various event orderings. We characterize simulation performance in three layers: physical system, simulation model, and simulator implementation; and focus on three performance measures: time, space, and event dependence. To compare and quantify the degree of event dependency of event orderings, we propose a relation called stricter and a measure called strictness, respectively. The proposed framework provides a framework to compare simulation performance across layers and allows the performance of different event orderings to be evaluated independent of implementation overheads from the physical system to its simulator implementation.